Meta unveils new LLaMA 4 AI models with enhanced capabilities

(Web Desk): Meta has introduced its new artificial intelligence (AI) model, LLaMA 4, to users. This new model will be available in four versions: LLaMA 4 Scout, LLaMA 4 Maverick, LLaMA 4 Behemoth, and LLaMA 4.

In a statement, Meta mentioned that all versions have been trained on data that includes text, images, and videos to better understand visual content. The success of Chinese company DeepMind's OpenAI models pushed Meta to develop a new model. DeepMind's AI models outperformed Meta's previous LLaMA models, prompting the social media company to accelerate the development of the new model.

According to Meta, LLaMA 4 Scout and Maverick are available for everyone on llama.com, while the training process for LLaMA 4 Behemoth is still ongoing. The company also stated that its AI assistant, Meta AI, which is integrated into WhatsApp, Messenger, Facebook, and Instagram, is being updated to LLaMA 4 in 40 countries. Currently, multi-modal features are limited to the United States but will be gradually expanded to other countries.

Meta mentioned in a blog post that LLaMA 4 models will mark the beginning of a new era and that this is just the start of the LLaMA 4 collection. Furthermore, Meta revealed that internal testing has shown that Maverick performs best as a general assistant and for chat, making it suitable for creative writing. In comparison to specific coding and other fields, Maverick can surpass OpenAI's GPT-4 and Google's Gemini 2.0. On the other hand, LLaMA 4 Scout has proven to be more effective in document summarization and logical reasoning.